The objective of this career proposal is to derive a consensus theoretical framework for distributed multi-vehicle cooperative control. In particular, the proposal focuses on distributed multi-vehicle formation clustering and rigid body attitude synchronization problems. While consensus algorithms start to emerge in cooperative control, they are usually simplified or ideal without accounting for the challenges in cooperative control problems. Therefore, their applicability to the above two problems is very limited.
The research plan of this career proposal consists of three thrusts: (i) analysis and design of novel, provably correct consensus algorithms and derivation of a unified consensus architecture for distributed multi-vehicle formation clustering problems, (ii) analysis and design of novel, provably correct consensus algorithms for distributed rigid body attitude synchronization problems, and (iii) experimental validation of the distributed algorithms on a multi-vehicle testbed. The distributed algorithms derived in the project will address consensus tracking of a reference state with inherent dynamics, consensus under actuator saturation, consensus in the absence of absolute or relative state derivative measurements, and consensus of coupled nonlinear systems modeled by rigid body attitude dynamics or Euler-Lagrange equations.
The research has significant impact on applications including space-based interferometry, environment monitoring, border patrol, and search and rescue. In collaboration with the Utah Water Research Laboratory, the PI plans to apply the research results to high-resolution remote sensing of land surface hydrologic processes by a team of unmanned aerial vehicles. The research also has impact on the fields of biology, economics, physics, and computer science, where the consensus/synchronization phenomenon is ubiquitous. The research plan will be integrated into an education plan that includes four elements, namely, undergraduate and graduate mentoring, curriculum development, outreach to high schools, and broad dissemination of the research results.
Recent advances in the miniaturization of computing, communication, sensing, and actuation have made it feasible to envision large numbers of autonomous vehicles (air, ground, and water) working cooperatively to accomplish an objective. Cooperative control of multiple autonomous vehicles has potential impact in numerous civilian, homeland security, and military applications. However, for all of these applications, communication bandwidth and power constraints will preclude centralized command and control. In this NSF funded research, innovative distributed consensus algorithms are derived for multi-vehicle cooperative control with limited communication bandwidth and power constraints. These distributed consensus algorithms guarantee that collective group behavior is achieved through local interaction with focuses on distributed multi-vehicle formation clustering and spacecraft attitude synchronization problems. While distributed algorithms have started to emerge in cooperative control, they are usually simplified or ideal without accounting for the challenges in cooperative control problems and hence with limited applicability. This research address several challenges in cooperative control including consensus tracking of a reference state with inherent dynamics, consensus under actuator saturation, consensus with reduced communication/sensing, and synchronization of coupled nonlinear systems modeled by rigid body attitude dynamics or Euler-Lagrange equations. The key outcomes include designing and analyzing various algorithms for agents with various dynamics including first-, second-, or high-order linear dynamics, rigid body attitude dynamics, and Euler-Lagrange dynamics, studying various real-world issues including sampled-data settings, optimality constraints, and time delay effects, and exploring collective period motions generated through Cartesian coordinate coupling and coupled harmonic oscillators, and developing distributed coordinated tracking algorithms with a dynamic leader in the presence of reduced interaction and partial measurements. The distributed consensus algorithms derived in the project are also experimentally validated on a multi-vehicle testbed consisting of a team of networked unmanned ground and aerial vehicles. In terms of intellectual merit, the innovative distributed consensus algorithms derived in the project enable that collective group behavior is achieved among multiple autonomous vehicles through local interaction despite the lack of centralized leadership and the existence of a dynamic, sparse, or intermittent network topology. In terms of broader impacts, the research has impact on applications including environment monitoring, border patrol, search and rescue, and space-based interferometry. The project has educational activities including robotics competition and outreach in a state-wide engineering state program for high school students.